I published this for readers on Medium as it is a big problem there
Dear friends, a few days ago, I published a comprehensive analysis on Medium.com about the shortcomings of glorified DeepSeek, but it struggled to reach its audience while sensationalized, one-sided posts were amplified and went viral, confusing the public. Some praised it so much, neglecting its hallucinations, saying it was from another planet. I responded by saying it was from China.
Fortunately, many readers discovered it through other platforms, and it even caught the attention of policymakers in several countries, influencing timely decisions on this critical issue.
I surveyed my email subscribers and found that many of them rely on AI tools for health advice as they think these tools are convenient, and they believe these AI tools are smart. The reality is that these tools, especially DeepSeek, are immature, and the public data they use for training might be wrong and usually obsolete. As DeepSeek has no web access, it is an extra risk.
While using reliable AI models with real-time web access and trusted sources like PubMed for preliminary research is reasonable, tools that lack updated information and generate responses based solely on outdated training data pose serious risks — even potential dangers.
When it comes to health and wellness, accuracy is vital. The wrong advice can lead to harmful choices, unnecessary stress, or even serious health risks. That is why I feel compelled to warn you about something that may be affecting more people than we realize — the misinformation risks of AI-generated health advice.
Recently, I conducted a deep investigation into DeepSeek with a team of testers in a government organization, an AI tool that has been gaining attention as an alternative to ChatGPT, Gemini, Copilot, or Claude.
While some have praised its capabilities on social media, my findings, like insights from experts, have shown serious flaws that should concern anyone who cares about making informed decisions in any important matter like health, well-being, nutrition, finance, and others.
After multiple tests, I found the following shortcoming:
DeepSeek’s training data only goes up to October 2023, and it can’t access new medical research or up-to-date guidelines. This means it may offer outdated, incomplete, or even dangerous recommendations, especially in health, where discoveries happen almost daily.
Unlike human doctors or trained medical professionals, AI should naturally learn from errors and refine responses as intended. However, DeepSeek repeated the same incorrect information multiple times during my testing for very basic inquiries, even after corrections were provided.
AI models sometimes generate convincing but false health claims. This is known as “hallucination” in AI research, which is part of my study. If you ask DeepSeek about a new treatment, a supplement, or a medical condition, there is no guarantee that the response is grounded in science, clinical trials, or expert consensus. It is very good at fabricating with boldness.
DeepSeek is hosted on servers in China, which means any health-related questions you ask could be stored, analyzed, or even used in ways you do not intend. Personal health concerns should be handled with care, and sharing sensitive medical information with an AI of uncertain data policies is a risk no one should take lightly.
That’s the reason several countries like Italy, Australia, South Korea, and some states in the US, like Texas, banned its use in their devices and warned the public, as I documented in my recent report.
I am not against AI. I believe AI can be a powerful tool in research and education. But regarding something as personal and essential as your health, you need verified sources, licensed professionals, and evidence-based information from reliable sources.
It is okay to do some preliminary research, but when tempted to ask an AI about a diagnosis, medication, diet, or treatment plan, stop and ask yourself: Would I trust their answer with my health and well-being?
For the most reliable health insights, I encourage you to check peer-reviewed medical research indexed in PubMed from peer-reviewed journals like JAMA, The Lancet, and Nature, licensed healthcare professionals such as family doctors, specialists, dietitians, psychotherapists, and trusted health organizations like WHO or CDC.
Be Critical, Be Cautious, Be Informed!
Technology is evolving fast, faster than our nature. The way we search for information is changing rapidly. But no matter how advanced AI becomes, our health deserves more than machine-generated answers.
An even greater risk is emerging. Some new writers copy and paste DeepSeek’s responses, rewording them with humanizers to sound more human and submitting them as original articles to various publications.
Since some AI detectors produce false negatives, these articles often go unnoticed. However, our publication has editors with health backgrounds who have identified and rejected hundreds of such misleading stories. So, please also be careful with what you read and hear on social media.
Please feel free to share this info with friends and family and encourage them to refrain from attempts to use AI tools for health advice because misinformation spreads fast and can be dangerous. Awareness is our best defense. Stay well, stay informed, and always choose scientific inquiry over speculation.
If you have had confusing experiences with AI tools and have used DeepSeek lately, please share your thoughts and experiences in the comment section. We can learn from each other’s experiences and insights.
Thank you for reading my newsletter. I wish you a healthy and happy life.
I write about many topics, including technology, health, and content marketing strategy. However, as a researcher in the cognitive science discipline, I wrote many guiding stories to inform and inspire you to improve your cognition and mental health.
Brain Health and Cognitive Function
I invite you to join my newsletters, where I offer experience-based content on health, content strategy, and technology topics to inform and inspire my readers.
Health and Wellness Network
Content Strategy, Development, & Marketing Insights
Technology Excellence and Leadership
